Analog parameter test system based on digital test equipment

Abstract

The invention discloses a simulation parameter test system based on digital test equipment, and relates to the field of simulation test, and the system comprises a verification control module which is used for obtaining the operation of specified test equipment and the obtaining authority of setting parameters, and sending a control instruction to each function module; the rule screening module is used for establishing a database, storing testable item data of the test equipment, pre-defining screening rules of testable items of the test equipment, screening corresponding data based on the rules and performing classification processing; the integrity calculation module is used for setting various parameter indexes for each test item, calculating the integrity of the various parameters under the current test item, and counting data missing in the indexes; according to the method, high-sensitivity parameters with high test item values are identified and supplemented preferentially, so that the prediction efficiency is ensured, interference of low-value parameters on the model is avoided, factors which have the greatest influence on a test result are identified, potential risks are pre-judged in advance, and the direction and amplitude of missing parameters are automatically adjusted.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of analog test technology, in particular to an analog parameter test system based on digital test equipment. BACKGROUND

[0002] In the process of electronic product research and development, testing and quality control, test equipment often involves a large number of complex parameters and data, and the absence or inaccuracy of these parameters directly affects the reliability and effectiveness of test results. With the improvement of semiconductor process precision and the popularization of test scenarios, traditional test methods face challenges such as large parameter system, dynamic changes in data gaps, low test efficiency, etc. In the prior art, due to the high actual test cost, analog testing is often affected by data loss, and it is difficult to obtain reliable supplementary prediction when data is missing. The prior art cannot accurately locate the correlation loss between multi-dimensional parameters, and cannot effectively predict the key parameter threshold when the data is incomplete. It is difficult to effectively identify the direction and amplitude of the data gap, and it is impossible to correct the preset indicators according to the data trend in the simulation results, and it is impossible to determine which parameter loss has a greater impact on the system test, resulting in insufficient test model reliability, and the test strategy cannot be adjusted in time. Intelligent test scenarios require the system to have self-correction capability to adapt to equipment aging, environmental drift and other factors, and need to combine historical data modeling and real-time trend analysis to achieve dynamic optimization. SUMMARY

[0003] (I) Technical problems solved In view of the above-mentioned shortcomings of the prior art, the present application provides an analog parameter test system based on digital test equipment, which can effectively solve the problems of the prior art.

[0004] (II) Technical solutions To achieve the above purpose, the present application is realized by the following technical solutions, The present application discloses an analog parameter test system based on digital test equipment, comprising: A verification control module is used to obtain the acquisition authority of the running and setting parameters of the specified test equipment, and to send control instructions to each functional module. A rule screening module is used to establish a database, store test equipment measurable item data, predefine test equipment measurable item screening rules, and screen and classify corresponding data based on the rules. A complete calculation module is used to set various parameter indicators for each test item, calculate the completeness of various parameters under the current test item, and count the missing data of the indicators. A missing analysis module is used to receive data from the complete calculation module, locate data gaps using matrix analysis method, and identify the direction and amplitude of the missing parameters. The correction model module is configured to establish a correction model by a time series network algorithm, the correction model is based on current available parameter data, and predicts a numerical range of a missing parameter, and a supplemented parameter set is formed according to the predicted numerical range; The simulation analysis unit is configured to construct a Verilog-A model of current test equipment behavior based on the supplemented parameter set, to obtain a change trend of a key indicator of a current test project within a specific period of time, to identify a key influencing factor, and to output weights of several key indicators; The simulation output module is configured to input the trend analysis result and the preset indicator weight into the Verilog-A model, to simulate performances of indicators of each test project within a specific period of time, to calculate a peak value, a mean value and a fluctuation range, and to output result data of the simulation operation and a corresponding evaluation report. The optimization unit is configured to evaluate a contribution degree of the supplemented prediction data in a final simulation test result, to adjust a missing indicator direction and a missing amplitude in current state data, and to correct according to a preset scale.

[0005] Further, the simulation analysis unit is deployed with a sub-module, the sub-module includes a test model module, a trend analysis module and a weight allocation module, the test model module and the trend analysis module are connected through a wireless network, and the trend analysis module and the weight allocation module are connected through a wireless network, and wherein: The test model module is configured to construct a basic Verilog-A model for simulation test according to attribute information and historical test case data of a measurable project of test equipment, to input the supplemented parameter set processed by the supplemented correction model into the basic Verilog-A model, to define key project indicators within a specific test period, and to construct a corresponding test scene for simulation operation; The trend analysis module is configured to perform time series trend analysis on historical data and current test data, and to output a change trend of each indicator within a specific period of time; The weight allocation module is configured to identify key factors affecting the performance of a measurement indicator in the trend analysis result, to classify the influencing factors into different factor categories according to the nature of the influencing factors, and to calculate weights of several key indicators of a current test project based on the classification result.

[0006] Further, the process of calculating the weights of the several key indicators of the current test project by the weight allocation module includes: Step a: based on the trend analysis result, extracting change characteristics of each test indicator within a specific period of time, identifying key factors having a significant influence on the performance of the indicator according to a mapping relationship between the characteristics and a pre-defined factor category, and classifying the key factors into corresponding factor categories according to physical properties, time series correlation and equipment parameter types; Step b: for each factor category, combine the contribution degree distribution of the same factor in historical data to establish a weight distribution matrix; Step c: process the weight distribution matrix through a normalization algorithm to generate an initial weight set, and introduce the current test environment constraint condition for dynamic adjustment; Step d: input the calculated weight into the Verilog-A model for simulation verification, if the deviation of the index performance in the simulation result from the preset target exceeds the threshold value, then re-execute steps a to c until the weight set converges to a stable state.

[0007] Further, the pre-defined factor categories in step a include device inherent parameters, environmental noise parameters and test timing parameters, and the Euclidean distance algorithm is used to match the characteristic vector and the category template when classifying.

[0008] Further, the weight distribution matrix in step b includes the sensitivity coefficient of each factor category to the test index, the timing influence intensity and the physical constraint condition, the weight distribution matrix determines the sensitivity coefficient through covariance analysis, the timing influence intensity is calculated by an autoregressive model, and the physical constraint condition is generated by mapping the device parameter threshold library.

[0009] Further, in the process of locating the data gap by the missing data analysis module, the parameter index integrity data under each test item is mapped into a two-dimensional matrix, wherein the row vector represents the parameter category, the column vector represents the test item, and the matrix element value represents the existence state and quantitative integrity of the corresponding parameter in the specific test item; The principal components of the matrix are extracted by singular value decomposition, the projection distribution of the parameter missing direction in the characteristic vector space is identified, the column missing density and the row missing density of each parameter category are calculated, and based on the joint analysis of the projection distribution and the missing density-intensity, the gap direction vector is established; According to the matrix sparsity index, the overall missing proportion is calculated, and the missing amplitude evaluation value with direction weight is generated combined with the gap direction vector module length.

[0010] Further, the construction process of the correction model in the correction model module is: A parameter prediction architecture is constructed based on a timing network algorithm, a multi-layer neural network structure with timing correlation is established, the network structure includes an input layer, a hidden layer and an output layer, wherein the input layer receives the standardized data of the current available parameters, and the output layer generates the predicted value of the missing parameters; Extract the complete parameter set in the historical database as a training sample, divide the sample into a training set and a validation set according to the time sequence, optimize the network weight through a back propagation algorithm, so that the root mean square error of the predicted value output by the correction model and the actual historical data is minimized; A parameter sensitivity evaluation function is defined, a sensitivity coefficient is calculated based on the response gradient of each parameter to the test item, and a sensitivity ranking list is generated, the function satisfying: ; Wherein is the sensitivity of the ith parameter, is the test item index, is the current parameter value, is the parameter standard deviation, is the index mean value; A threshold constraint condition is set to truncate the parameter value range generated by prediction; According to the sensitivity ranking list, the missing parameters are assigned priorities, and the parameters with a sensitivity higher than a preset threshold are preferentially supplemented to the parameter set to generate a supplemented parameter set containing predicted values.

[0011] Further, the optimization unit is deployed with a sub-module, the sub-module includes a contribution evaluation module, a dynamic adjustment module and a configuration application module, and the contribution evaluation module is interactively connected with the dynamic adjustment module and the configuration application module through a wireless network, wherein: The contribution evaluation module is used to evaluate the contribution of the supplemented prediction data to the final simulation test result, and when the contribution is lower than a preset threshold, a data correction reminder is triggered to compare the deviation between the supplemented parameter set and the preset target data; The dynamic adjustment module is used to obtain the prediction data contribution evaluation result, and when the contribution evaluation module judges that the prediction data contribution is lower than the preset threshold, the missing index direction and the missing amplitude in the current state data are adjusted, and the correction is performed according to a preset scale; The configuration application module is used to re-input the state data adjusted by the dynamic adjustment module into the correction model module, and perform the next round of prediction through the correction model.

[0012] Further, the correction logic of the missing index direction and the missing amplitude in the dynamic adjustment module is: The projection component of the gap direction vector in each feature vector space is obtained, the amplitude and polarity of the projection component are extracted, the main direction weight of the parameter missing is obtained, the missing correlation strength of each parameter category in the continuous test item is output in combination with the column missing density and the row missing density; According to the product of the matrix sparsity index and the gap direction vector module length, a direction-weighted missing amplitude coefficient is generated, and the preset scale is defined as a combination function of the dynamic adjustment step and the direction weight factor; Based on the direction weight factor, the direction vector of the missing parameter is normalized, the adjustment step is distributed to the missing direction of the corresponding parameter category according to the weight, and a correction amount with direction bias is generated; According to the preset threshold value constraint correction amount amplitude range, if the correction amount exceeds the threshold value, it is compressed to the maximum allowed value in proportion; The modified direction vector and amplitude value are superimposed to the missing position of the current state data, the gap direction vector module length and sparsity index are updated, and the iteration is performed until the contribution reaches the preset threshold.

[0013] Further, the verification control module is interactively connected with the rule screening module and the complete calculation module through a wireless network, the missing analysis module is interactively connected with the complete calculation module, the correction model module and the simulation analysis unit through a wireless network, and the simulation output module is interactively connected with the simulation analysis unit and the optimization unit through a wireless network.

[0014] (Three) beneficial effects Compared with the known prior art, the technical scheme provided by the application has the following beneficial effects, By automatically screening and classifying data through test item attributes, quickly positioning parameter completeness gap, accurately identifying the direction and amplitude of the missing index, establishing a correction model with time sequence correlation, identifying and prioritizing high-sensitive parameters with higher test item value through sensitivity sorting and physical threshold limitation, and then avoiding the interference of low-value parameters on the model while ensuring the prediction efficiency, the effectiveness of the supplemented parameter set is improved, the priority of the missing parameter is valued, and the efficient correction of the supplemented prediction data is realized, ensuring the continuous optimization of the prediction value under the physical constraint condition.

[0015] Through the simulation results output by the simulation analysis unit based on the Verilog-A model, the periodical index is defined and the trend is analyzed, the weight of the key influence factor is quantified, the factor that has the greatest impact on the test result is identified, the system can further simulate the future test scene, predict potential risks in advance, dynamically adjust the index weight based on the influence factor classification, make the test model more suitable for actual working conditions, avoid the deviation caused by static weight, and improve the test adaptability in complex scenes.

[0016] The contribution threshold is introduced by the optimization unit to judge the effectiveness of the prediction data, the projection component of the gap direction vector and the missing correlation strength are defined to dynamically adjust the step size and direction weight factor, the missing data is updated through threshold compression and iteration superposition until the contribution meets the standard, if the supplemented data is insufficient to improve the trend analysis, the missing parameter direction and amplitude are automatically adjusted, the prediction value is regenerated, the correctness of data correction is improved, and the manual intervention is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description only show some embodiments of the present application, and based on these drawings, other drawings can be obtained by one of ordinary skill in the art without any creative effort.

[0018] Figure 1 The figure is a schematic diagram of the overall framework of the present application. Figure 2 The figure is a schematic diagram of the framework of the simulation analysis unit in the present application. Figure 3 The figure is a schematic diagram of the framework of the optimization unit in the present application.

[0019] The numbers in the figure respectively represent 1, a verification control module; 2, a rule screening module; 3, a complete calculation module; 4, a missing analysis module; 5, a correction model module; 6, a simulation analysis unit; 61, a test model module; 62, a trend analysis module; 63, a weight distribution module; 7, a simulation output module; 8, an optimization unit; 81, a contribution evaluation module; 82, a dynamic adjustment module; and 83, a configuration application module. DETAILED DESCRIPTION

[0020] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by one of ordinary skill in the art without any creative effort belong to the scope of protection of the present application.

[0021] The present application will be further described below in combination with embodiments.

[0022] Embodiment 1 The simulation parameter test system based on digital test equipment in this embodiment, as shown in Figure 1 Figure 3 includes: A verification control module 1 is used to acquire the acquisition authority of the running and setting parameters of the specified test equipment, and send control instructions to each functional module. A rule screening module 2 is used to establish a database, store the testable item data of the test equipment, predefine the screening rules of the testable items of the test equipment, and process the corresponding data based on the rule screening and classification. In the classification processing of the rule screening module 2, the data is classified according to the function category, the test mode category and the historical test record category. ​The complete calculation module 3 is used for setting various parameter indexes for each test item, calculating the completeness of various parameters under the current test item, and counting the missing data of the indexes; The missing analysis module 4 is used for receiving the data submitted by the complete calculation module 3, positioning the data gap by using the matrix analysis method, identifying the missing parameter direction and amplitude, and mapping the parameter index completeness data under each test item into a two-dimensional matrix in the process of positioning the data gap by the missing analysis module 4, wherein the row vector represents the parameter category, the column vector represents the test item, and the matrix element value represents the existence state and quantitative completeness of the corresponding parameter in the specific test item; The main components of the matrix are extracted by singular value decomposition, the projection distribution of the parameter missing direction in the eigenvector space is identified, the column missing density (the number of missing items / the total number of test items) and the row missing density (the number of continuous missing test items / the total number of test items) of each parameter category are calculated, and the gap direction vector is established based on the joint analysis of the projection distribution and the missing density-intensity; According to the matrix sparsity index (1-non-empty element number / total matrix element number), the overall missing proportion is calculated, and the missing amplitude evaluation value with direction weight is generated in combination with the gap direction vector module length; By constructing the parameter and test item matrix and fusing singular value decomposition and multi-dimensional missing quantization algorithm, the limitations of traditional single statistics or local interpolation method are broken through. The prior art usually only performs linear analysis based on the discrete parameter missing rate, and it is difficult to capture the correlation gap mode between the parameters and the test items. The direction vector projection analysis and dynamic density and intensity joint calculation are combined in the scheme, not only the systematic parameter missing direction across the test items is accurately positioned, but also the spatial distribution characteristics of the missing amplitude are quantified through the gap direction vector module length, the recognition ability of the implicit data gap is significantly improved, the structured missing evaluation data with spatial weight attribute is provided for the subsequent correction model, and the error accumulation problem caused by ignoring the topological relationship between the parameters in the prior art is solved; The correction model module 5 is used for establishing a correction model by a time series network algorithm, training the correction model according to the historical data as a training sample, predicting the numerical range of the missing parameter based on the current available parameter data, forming a supplemented parameter set according to the predicted numerical range, preferentially supplementing the parameters with high sensitivity to the test item, and limiting the supplementary value within the physical reasonable threshold; The construction process of the correction model is as follows: The parameter prediction architecture is constructed based on the time series network algorithm, a multi-layer neural network structure with time series correlation is established, the network structure includes an input layer, a hidden layer and an output layer, the input layer receives the standardized data of the current available parameters, and the output layer generates the predicted value of the missing parameter; Extract the complete parameter set from the historical database as training samples, divide the samples into training set and validation set according to the time series, and optimize the network weights through the back propagation algorithm to minimize the root mean square error between the predicted value output by the modified model and the actual historical data; Define the parameter sensitivity evaluation function, calculate the sensitivity coefficient based on the response gradient of the test item to each parameter, and generate a sensitivity ranking list. The function satisfies: ; in is the sensitivity of the i-th parameter, For test project indicators, is the current parameter value, is the parameter standard deviation, is the mean value of the indicator; Set threshold constraints to truncate the range of parameter values ​​generated by the prediction; Prioritize the missing parameters according to the sensitivity ranking list, and prioritize the parameters with a sensitivity higher than the preset threshold to add to the parameter set to generate a supplemented parameter set containing the predicted value; The simulation analysis unit 6 is used to construct a Verilog-A model of the current test equipment behavior based on the supplemented parameter set, obtain the change trend of the key indicators pre-defined in the current test project within a specific period, identify key influencing factors, output the weights of several key indicators, embed test environment constraints, support user-defined test indicators and generate adapted test vectors; The simulation analysis unit 6 is deployed with submodules, including: a test model module 61, a trend analysis module 62 and a weight allocation module 63. The test model module 61 and the trend analysis module 62 are interactively connected via a wireless network, and the trend analysis module 62 and the weight allocation module 63 are interactively connected via a wireless network, wherein: The test model module 61 is used to construct a basic Verilog-A model for simulation testing based on the attribute information of the test equipment's measurable items and historical test case data, input the supplemented parameter set processed by the supplemented and corrected model into the basic Verilog-A model, define key project indicators within a specific test period, and construct corresponding test scenarios for simulation operation; The trend analysis module 62 is used to perform time series trend analysis on historical data and current test data, and output the changing trend of each indicator within a specific period of time; The weight allocation module 63 is used to identify key factors that affect the performance of measurement indicators in the trend analysis results, classify the influencing factors into different factor categories according to their nature, and calculate the weights of several key indicators under the current test item based on the classification results; The simulation output module 7 is configured to input the trend analysis result and the preset index weight into the Verilog-A model, simulate the performance of each test item index in a specific period, calculate the peak value, the mean value and the fluctuation range, and output the result data of the simulation operation and the corresponding evaluation report. The optimization unit 8 is configured to evaluate the contribution of the supplementary prediction data in the final simulation test result, adjust the missing index direction and the missing amplitude in the current state data, and correct the missing index direction and the missing amplitude according to the preset scale. The optimization unit 8 is deployed with a sub-module, and the sub-module includes a contribution evaluation module 81, a dynamic adjustment module 82 and a configuration application module 83. The contribution evaluation module 81 is connected to the dynamic adjustment module 82 and the configuration application module 83 through a wireless network. The contribution evaluation module 81 is configured to evaluate the contribution of the supplementary prediction data in the final simulation test result, and trigger a data correction reminder when the contribution is lower than a preset threshold, and compare the deviation between the parameter set after the supplement and the preset target data. The dynamic adjustment module 82 is configured to obtain the contribution evaluation result of the prediction data, and adjust the missing index direction and the missing amplitude in the current state data when the contribution evaluation module 81 judges that the contribution of the prediction data is lower than the preset threshold. The configuration application module 83 is configured to input the state data adjusted by the dynamic adjustment module 82 into the correction model module 5 as the input of the next round of prediction. The verification control module 1 is connected to the rule screening module 2 and the complete calculation module 3 through a wireless network, the missing analysis module 4 is connected to the complete calculation module 3, the correction model module 5 and the simulation analysis unit 6 through a wireless network, and the simulation output module 7 is connected to the simulation analysis unit 6 and the optimization unit 8 through a wireless network.

[0023] Compared with the prior art, the test data is classified according to the function category, the test mode and the historical test record, and the matrix singular value decomposition method is used to accurately locate the parameter gap, so that the quantitative evaluation of the data completeness is realized, the missing amplitude evaluation value with the direction weight is generated by combining the gap direction vector and the matrix sparsity index, and the missing parameter is accurately predicted and corrected under the premise of ensuring the rationality of the data. The modified model of the multi-layer neural network is constructed by using the time sequence network algorithm. On the basis of optimizing the weight by back propagation, the parameter value range can be effectively truncated, and the parameter sensitivity can be sorted according to the response gradient of the test item, so that the parameters with high sensitivity to the test item are preferentially supplemented. Through the multi-level strategy of Verilog-A model construction, trend analysis, key factor identification and weight distribution, the key indicator performance of each test item in a specific period of time can be simulated and evaluated in real time and accurately, which significantly improves the completeness, reliability of the data and the prediction accuracy of the test indicators, and has higher self-adaptability, prediction accuracy and operation efficiency compared with the prior art.

[0024] ②Embodiment 2 On the other hand, the embodiment also provides another optimization mechanism based on embodiment 1, which is a scheme for calculating the weight of several key indicators under the current test item. The specific process includes: Step a: based on the trend analysis result, the change characteristics of each test indicator in a specific period of time are extracted, the key factors that have a significant influence on the indicator performance are identified according to the mapping relationship between the characteristics and the predefined factor categories, and they are classified into corresponding factor categories according to physical properties, time sequence correlation and device parameter types. The predefined factor categories include device inherent parameters, environmental noise parameters and test timing parameters, and the Euclidean distance algorithm is used to match the characteristic vector and the category template when classifying; Step b: for each factor category, a weight distribution matrix is established by combining the contribution degree distribution of the same type of factor in the historical data; the weight distribution matrix includes the sensitivity coefficient, the time sequence influence intensity and the physical constraint condition of each factor category to the test indicator, the sensitivity coefficient is determined by covariance analysis, the time sequence influence intensity is calculated by the autoregressive model, and the physical constraint condition is generated by mapping the device parameter threshold library; Step c: the weight distribution matrix is processed by a normalization algorithm to generate an initial weight set, and the current test environment constraint condition is introduced for dynamic adjustment; to ensure that the weight of each key indicator satisfies =1, and ∈[0,1], wherein represents the maximum weight value of the i-th indicator; the dynamic adjustment adopts the Lagrange multiplier method to optimize the initial weight under the boundary condition of the environmental constraint; Step d: the calculated weight is input into the Verilog-A model for simulation verification, if the deviation of the indicator performance from the preset target in the simulation result exceeds the threshold value, steps a to c are re-executed until the weight set converges to a stable state.

[0025] In the embodiment, by extracting features from the trend analysis results and using the Euclidean distance algorithm to realize accurate factor matching, the significant influencing factors of each test index are scientifically classified according to the equipment inherent parameters, environmental noise parameters and test timing parameters, thereby optimizing the identification process of the key index influencing factors, fully utilizing the contribution degree information of each factor in the historical data, and combining the mapping of the physical parameter threshold library to generate physical constraint conditions, so that the weight calculation is more in line with the actual test environment and equipment properties. The calculated weight is fed back to the Verilog-A model for simulation verification, and when the deviation exceeds the threshold, it is automatically iteratively optimized, significantly improving the accuracy of the simulation results and the robustness of the system as a whole. Compared with the prior art, the embodiment has obvious advantages in improving the prediction accuracy of key indicators, dynamic adaptive ability and data processing efficiency.

[0026] ③Embodiment 3 The embodiment provides a correction logic for missing index direction and missing amplitude, specifically: Obtain the projection component of the gap direction vector in each feature vector space, extract the amplitude and polarity of the projection component, obtain the main direction weight of the parameter missing, combine the column missing density and the row missing density, and output the missing correlation strength of each parameter category in the continuous test project; According to the product of the matrix sparsity index and the gap direction vector length, generate the direction-weighted missing amplitude coefficient, define the preset scale as a combination function of the dynamic adjustment step and the direction weight factor, adjustment step = missing amplitude coefficient x log (1 + missing correlation strength); direction weight factor = projection component amplitude / ∑ projection component amplitude x polarity coefficient; Based on the direction weight factor, normalize the direction vector of the missing parameter, distribute the adjustment step to the missing direction of the corresponding parameter category according to the weight, and generate a correction amount with direction bias; According to the amplitude range of the preset threshold constraint correction amount, if the correction amount exceeds the threshold, it is compressed to the maximum allowed value in proportion, to ensure the physical feasibility of the supplementary prediction value; Superimpose the corrected direction vector and amplitude value to the missing position of the current state data, update the gap direction vector length and the sparsity index, and iterate until the contribution degree reaches the preset threshold.

[0027] The embodiment dynamically adjusts the directionality and amplitude of the missing index, which not only accurately captures the multi-dimensional missing information in the test data, but also takes into account the physical constraint conditions in the correction process, realizes efficient, fine and adaptive data completion, and has obvious advantages in missing correction accuracy, model robustness and iterative optimization efficiency compared with the prior art.

[0028] Working principle: the system in the application obtains the control authority of the test equipment by the verification control module 1 when being carried, the rule screening module 2 filters the corresponding available data in the database based on the testable items of the test equipment, classifies, calculates the completeness of each category parameter on a certain test item through the complete calculation module 3, analyzes the state data of the missing index direction and amplitude of the current completeness through the missing analysis module 4; the correction model module 5 constructs the supplementary correction model based on the original parameter missing state data, takes the historical data as the reference, predicts the data supplement based on the missing part of the current available parameter, the test model module 61 constructs the simulation test model based on the testable item attribute of the test equipment and its historical test case data, inputs the supplemented parameter, the trend analysis module 62 defines the required project index in a specific period, performs trend analysis based on the defined index, and according to the trend analysis result, the weight distribution module 63 classifies based on the influence factor in the trend analysis, calculates the weight of a plurality of indexes under the current test item based on the classification result, based on the trend analysis result and a plurality of preset index weights, the simulation output module 7 extracts the trained test model to simulate running, and outputs a project data performance in a specific period; The contribution evaluation module 81 evaluates the contribution degree of the prediction supplement data based on the trend analysis result, when the contribution degree is insufficient, the dynamic adjustment module 82 adjusts the missing index direction and amplitude of the state data according to the preset scale, the configuration application module 83 re-submits the adjusted state data to the missing analysis module 4 and the correction model module 5 based on the adjusted state data, and performs the next round of prediction; The system solves the core pain points of low data reliability and poor model adaptability in complex test environment by constructing parameter integrity evaluation, matrix gap positioning and prediction correction closed loop, and meets the urgent needs of high-precision digital equipment in research and development verification, mass production testing and other links for automatic and intelligent test system.

[0029] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements will not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. The analog parameter test system based on digital test equipment is characterized by: include: A verification control module (1) is used to obtain the authority to operate and set parameters of a specified test device and to send control instructions to each functional module; A rule screening module (2) is used to establish a database, store data of various measurable items of the test equipment, predefine screening rules for the measurable items of the test equipment, screen the corresponding data based on the rules and classify and process them; The complete calculation module (3) is used to set various parameter indicators for each test item, calculate the completeness of each category of parameters under the current test item, and count the missing data of the indicators; The missing analysis module (4) is used to receive the data submitted by the complete calculation module (3), locate the data gap using matrix analysis method, and identify the direction and magnitude of the missing parameters; A correction model module (5) is used to establish a correction model through a time series network algorithm, wherein the correction model predicts the value range of the missing parameter based on the currently available parameter data, and forms a supplemented parameter set according to the predicted value range; A simulation analysis unit (6) is used to construct a Verilog-A model of the current test equipment behavior based on the supplemented parameter set, obtain the change trend of the key indicators pre-defined in the current test item within a specific period of time, identify key influencing factors, and output the weights of several key indicators; The simulation output module (7) is used to input the trend analysis results and the preset indicator weights into the Verilog-A model, simulate the performance of each test item indicator in a specific period of time, calculate the peak value, mean value and fluctuation range, and output the result data of the simulation operation and the corresponding evaluation report; The optimization unit (8) is used to evaluate the contribution of the supplementary prediction data to the final simulation test results, adjust the missing indicator direction and missing amplitude in the current state data, and make corrections according to the preset scale.

2. The analog parameter testing system based on digital testing equipment according to claim 1, characterized in that: The simulation analysis unit (6) is provided with submodules at the lower level, and the submodules include: a test model module (61), a trend analysis module (62) and a weight distribution module (63). The test model module (61) and the trend analysis module (62) are interactively connected via a wireless network, and the trend analysis module (62) and the weight distribution module (63) are interactively connected via a wireless network, wherein: A test model module (61) is used to construct a basic Verilog-A model for simulation testing based on the attribute information of the test equipment's measurable items and historical test case data, input the supplemented parameter set processed by the supplemented correction model into the basic Verilog-A model, define key project indicators within a specific test period, and construct a corresponding test scenario for simulation operation; A trend analysis module (62) is used to perform time series trend analysis on historical data and current test data, and output the changing trend of each indicator within a specific time period; The weight allocation module (63) is used to identify the key factors that affect the performance of the measurement indicators in the trend analysis results, classify the influencing factors into different factor categories according to their properties, and calculate the weights of several key indicators under the current test item based on the classification results.

3. The analog parameter testing system based on digital testing equipment according to claim 2, characterized in that: The process of the weight allocation module (63) calculating the weights of several key indicators under the current test item includes: Step a: Based on the trend analysis results, extract the change characteristics of each test indicator within a specific period of time, identify the key factors that have a significant impact on the indicator performance based on the mapping relationship between the characteristics and predefined factor categories, and classify them into corresponding factor categories according to physical properties, time series correlation, and device parameter type; Step b: For each factor category, combine the contribution distribution of similar factors in historical data to establish a weight distribution matrix; Step c: Processing the weight distribution matrix through a normalization algorithm to generate an initial weight set, and introducing the current test environment constraints for dynamic adjustment; Step d: Input the calculated weights into the Verilog-A model for simulation verification. If the deviation between the performance of the indicators in the simulation results and the preset targets exceeds the threshold, re-execute steps a to c until the weight set converges to a stable state.

4. The analog parameter testing system based on digital testing equipment according to claim 3, characterized in that: The predefined factor categories in step a include device intrinsic parameters, environmental noise parameters and test timing parameters, and the Euclidean distance algorithm is used to match the feature vector and the category template during classification.

5. The analog parameter testing system based on digital testing equipment according to claim 3, characterized in that: The weight distribution matrix in step b includes the sensitivity coefficient of each factor category to the test indicator, the timing impact intensity and the physical constraint conditions. The weight distribution matrix determines the sensitivity coefficient through covariance analysis, the timing impact intensity is calculated by the autoregressive model, and the physical constraint conditions are generated by mapping the device parameter threshold library.

6. The analog parameter testing system based on digital testing equipment according to claim 1, characterized in that: In the process of locating data gaps, the missing analysis module (4) maps the parameter indicator integrity data under each test item into a two-dimensional matrix, wherein the row vector represents the parameter category, the column vector represents the test item, and the matrix element value represents the existence state and quantitative integrity of the corresponding parameter in the specific test item; The principal components of the matrix are extracted through singular value decomposition, the projection distribution of the parameter missing direction in the eigenvector space is identified, the column missing density and row missing density of each parameter category are calculated, and the gap direction vector is established based on the joint analysis of the projection distribution and the missing density-intensity. The overall missing ratio is calculated based on the matrix sparsity index, and the missing amplitude evaluation value with directional weight is generated by combining the modulus of the gap direction vector.

7. The analog parameter testing system based on digital testing equipment according to claim 1, characterized in that: The construction process of the correction model in the correction model module (5) is as follows: Constructing a parameter prediction architecture based on a time series network algorithm, establishing a multi-layer neural network structure with time series correlation, the network structure includes an input layer, a hidden layer, and an output layer, wherein the input layer receives the standardized data of the currently available parameters, and the output layer generates the predicted values ​​of the missing parameters; Extract the complete parameter set from the historical database as training samples, divide the samples into training set and validation set according to the time series, and optimize the network weights through the back propagation algorithm to minimize the root mean square error between the predicted value output by the modified model and the actual historical data; Define a parameter sensitivity evaluation function, calculate the sensitivity coefficient based on the response gradient of the test item to each parameter, and generate a sensitivity ranking list. The function satisfies: ； in is the sensitivity of the i-th parameter, For test project indicators, is the current parameter value, is the parameter standard deviation, is the mean value of the indicator; Set threshold constraints to truncate the range of parameter values ​​generated by the prediction; The missing parameters are prioritized according to the sensitivity ranking list, and the parameters with sensitivity higher than the preset threshold are added to the parameter set first to generate a supplemented parameter set containing the predicted values.

8. The analog parameter testing system based on digital testing equipment according to claim 1, characterized in that: The optimization unit (8) is provided with submodules at a lower level, and the submodules include a contribution evaluation module (81), a dynamic adjustment module (82), and a configuration application module (83). The contribution evaluation module (81) is interactively connected to the dynamic adjustment module (82) and the configuration application module (83) via a wireless network, wherein: A contribution evaluation module (81) is used to evaluate the contribution of the supplementary prediction data to the final simulation test results. When the contribution is lower than a preset threshold, a data correction reminder is triggered to compare the deviation between the supplemented parameter set and the preset target data; A dynamic adjustment module (82) is used to obtain the prediction data contribution evaluation result, and when the contribution evaluation module (81) determines that the prediction data contribution is lower than a preset threshold, adjust the missing indicator direction and missing amplitude in the current state data and make corrections according to a preset scale; The configuration application module (83) is used to input the state data adjusted by the dynamic adjustment module (82) back into the correction model module (5) and perform the next round of prediction through the correction model.

9. The analog parameter testing system based on digital testing equipment according to claim 8, characterized in that: The correction logic of the missing indicator direction and missing magnitude in the dynamic adjustment module (82) is: Obtain the projection component of the gap direction vector in each eigenvector space, extract the amplitude and polarity of the projection component, obtain the main direction weight of the parameter missing, combine the column missing density and row missing density, and output the missing association strength of each parameter category in the continuous test items; Generate a direction-weighted missing amplitude coefficient based on the product of the matrix sparsity index and the modulus of the gap direction vector, and define the preset scale as a combination function of the dynamic adjustment step size and the direction weight factor; Based on the directional weight factor, the directional vector of the missing parameter is normalized, and the adjustment step size is distributed to the missing direction of the corresponding parameter category according to the weight, thereby generating a correction amount with a directional bias; The amplitude range of the correction amount is constrained according to the preset threshold value. If the correction amount exceeds the threshold value, it is proportionally compressed to the maximum allowable value; The corrected direction vector and amplitude value are superimposed on the missing position of the current state data, the modulus of the gap direction vector and the sparsity index are updated, and the iteration is performed until the contribution reaches the preset threshold.

10. The analog parameter testing system based on digital testing equipment according to claim 1, characterized in that: The verification control module (1) is interactively connected to the rule screening module (2) and the complete calculation module (3) via a wireless network; the missing analysis module (4) is interactively connected to the complete calculation module (3), the correction model module (5) and the simulation analysis unit (6) via a wireless network; and the simulation output module (7) is interactively connected to the simulation analysis unit (6) and the optimization unit (8) via a wireless network.

Images